RoboGraphBench - Indoor: leaderboard
Metric: Task success rate (%; episodes that reach the goal and stop correctly, averaged over the baseline and six intervention conditions; semantic closed-loop evaluation on partially observed symbolic scene graphs, at most 100 steps; 189 navigation-enabled indoor episodes from 27 scenes). Source: arxiv.org. Saturation forecast: Around December 2026. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 82.5 |
| 2 | Claude Opus 4.7 | 80.4 |
| 3 | GLM-5.2 | 57.1 |
| 4 | Gemini 3.1 Pro (Preview) | 42.3 |
| 5 | GPT-5.4 | 41.8 |
| 6 | Qwen 3.6 Plus | 28.6 |
| 7 | DeepSeek V4 Pro | 23.3 |
| 8 | Qwen 3.7 Plus | 21.7 |
| 9 | Qwen 3 VL 8B | 0.5 |
Interactive version: theaggregate.ai/benchmark?slug=robographbench-indoor · How It Works · Data refreshed daily, snapshot 2026-09-26.